96 Annals of the Carnegie Museum. 



notch, laterobasal angles, etc. Radial system strong, but ^•ariable; 

 always a transverse (lateral) radius on each side, directed a little 

 upwards; a pair of apical radii not fai apart, or sometimes only one; 

 sometimes two basal ladii; polygonal areas sometimes slightly 

 developed. Apical circuli longitudinal, but failing apically. This 

 is not far from the type of Leporinus megalepis. It is also strongly 

 suggestive of the African Petersius, although the fishes are very dis- 

 similar. Superficially the fish Anostomus looks like the African 

 Neohorus, which has totally different (ctenoid) scales. 



Prochilodin^. 

 Prochilodus rnhrotccniatus Schomburgk (Plate XXIV). Large scales, 

 about nine and one-half mm. long and ten mm. broad, the latero- 

 basal angles rounded, the basal middle emarginate, with or without 

 a radius running to nucleus; one to three paiis of lateral radii, 

 more or less joined, U-like at base, and a single apical radius or none, 

 nucleus a little apicad of middle, more or less multiple; apical margin 

 finely irregularly dentate; apical field (bounded by the uppermost 

 lateral radii) with coarse vertical (oblique toward middle) circuli 

 basally, but beyond this the circuli are entirely broken up to form a 

 dense labyrinthine pattern; lateral and basal circuli fine and 

 regular. A remarkable type of scale, representing an early stage 

 in the development of the ctenoid character, combined with an 

 Alestiform radial pattern. So far as the scales go, it must be con- 

 sidered a stem-form. 



CHILODIN.E. 



Chilodus punctatiis Miiller & Troschel (Plate XXV'I, fig. i) was 

 actually referred to Citharinus by Cuvier and Valenciennes, but 

 Boulenger places the related Ccenotropus in the Hemiodinae (Hemio- 

 dontinae). The scale is very distinctive. It has indeed the Curi- 

 matoid shape, but a strong transverse line (part of alestiform pat- 

 tern), and the apical circuli longitudinal, but not reaching the apex, 

 and hence the scale is not ctenoid. This doubtless illustrates the 

 beginning of the development which culminates in such specialized 

 ctenoid scales as those of the African Xenocharax. The fact that 

 in this and other genera South America supplies types connecting 

 the extremely different African groups typified by Distichodus and 

 Alestes, may be taken as an indication that the Characinidae origi- 

 nated in the Neotropical rather than the Ethiopian region. 



